Conway-Maxwell-Poisson Distribution

When parameters hit positive infinity, the gradients and HMC/optimization break down. If you really need this case, you can just use the Bernoulli.

OK, thanks for that. And I found no success in tweaking the while loop either.